2019-03-18 09:53:17 UTC - Christophe Bornet: Hi. Does Geo-replication preserve order from the replicated region ? And does it generate duplicates or is idempotency applied ? ---- 2019-03-18 09:57:03 UTC - Sijie Guo: > Does Geo-replication preserve order from the replicated region ?
yes > And does it generate duplicates or is idempotency applied ? I don’t think idempontent producing is enabled by default. although it should be straightforward to make it enable by default ---- 2019-03-18 10:12:00 UTC - Christophe Bornet: If I set deduplication for the global namespace, will it apply to the replication (I guess yes, but if you could confirm) ? ---- 2019-03-18 10:26:36 UTC - Alexandre DUVAL: Poke ---- 2019-03-18 12:58:02 UTC - Maarten Tielemans: We are still in the process of optimising our tail latencies, but have have some unclarity on the numbers in Prometheus/Grafana. We have write-count 3, ensemble 3 and ack 2 configured as defaults on the topic. For example, looking at the Add Entry Latency (99.99%), all 3 brokers report values above 100ms for the whole 15 minute sample range. This is NOT reflected in the output of pulsar-perf which sometimes has 99.99 latencies of 6ms and sometimes goes up to 100ms. Similarly, the Journal Sync Latency (Max) is above 140ms across the same 15 minute sample range. I can share access to the Grafana dashboard if desired. ---- 2019-03-18 13:50:23 UTC - Darragh: java11 wasn't in the repositories at all, so I installed the binaries myself and those did not have shenandoah ---- 2019-03-18 13:50:33 UTC - Darragh: the java8 binaries I was using do have shenandoah ---- 2019-03-18 13:50:43 UTC - Darragh: but the performance is similar if not worse ---- 2019-03-18 15:26:17 UTC - Maarten Tielemans: @Matteo Merli with you working on Streamlio, do you or your company provide professional services to investigate these sort of problems? at what rate? ---- 2019-03-18 16:41:18 UTC - Matteo Merli: Correct. The dedup feature is activated on the namespace. If it’s on, it will be applied both to local producer and to geo-replication ---- 2019-03-18 16:51:38 UTC - Christophe Bornet: Cool, thanks ! ---- 2019-03-18 18:00:12 UTC - Vikas: hi @David Kjerrumgaard how are you! I need to consume the data from the pulsar-log-topic and send it to Splunk. I am able to pull the data using the ConsumePulsar processor and but unable to send it using PutSplunk processor. We have Splunk on cloud and security has provided us with the `token auth` but I believe we have only SSL way of authentication available in the PutSplunk processor. Please advise. ---- 2019-03-18 18:03:21 UTC - David Kjerrumgaard: @Vikas You will need to reach out to the NiFi community to see if someone can add that authentication capability to the PutSplunk processor. ---- 2019-03-18 18:03:59 UTC - David Kjerrumgaard: Or you can make the changes yourself, and attempt to commit the change back into NiFi. ---- 2019-03-18 18:25:34 UTC - Vikas: sure thanks @David Kjerrumgaard ---- 2019-03-18 18:25:55 UTC - David Kjerrumgaard: If you need help let me know. ---- 2019-03-18 19:31:57 UTC - Ivan Kelly: @JAYARAM NAGARAJAN how many messages have you written in total? ---- 2019-03-18 20:26:09 UTC - Alexandre DUVAL: @Matteo Merli? ---- 2019-03-18 20:54:51 UTC - David Kjerrumgaard: This file doesn't show the "managedLedgerOffloadDriver=s3' property. Did you set that value in the standalone.conf file? ---- 2019-03-19 00:11:46 UTC - Victor Siu: @Victor Siu has joined the channel ---- 2019-03-19 01:36:50 UTC - Kenan Dalley: @Kenan Dalley has joined the channel ---- 2019-03-19 01:41:38 UTC - Kenan Dalley: Hi. I've got a question about Pulsar Functions. They're run on the Pulsar cluster, but what kinds of dependencies are allowed in the code? For example, if I want to connect to C* for a validation check with a python script, will that work? Or are they limited to only the Pulsar Client dependencies? ---- 2019-03-19 01:45:29 UTC - Sijie Guo: you can pull in your dependencies and package your code and dependencies into a zip file. ---- 2019-03-19 01:46:19 UTC - Kenan Dalley: oh, nice. ok, thanks! that's with v2.3.0, right? ---- 2019-03-19 01:50:56 UTC - Sijie Guo: yes that is 2.3.0 ---- 2019-03-19 01:52:34 UTC - Kenan Dalley: great, thanks! ---- 2019-03-19 07:44:31 UTC - Ingmar Bengtsson: @Ingmar Bengtsson has joined the channel ----
